Standdown AF2
august 2002
Reason for standdown:
During last run timeouts on the theta mechanism were causing problems
when starting the instrument up. Faultreports: 15006/15016, during the run
timeout period was made longer.
Place:
Test Focal station.
Work done:
It was decided to do a complete stripdown of the gripper unit. This has
never been serviced since it was put into operation in 97. This way a complete
inspection and necessary repairs could be caried out. Further it was an
oportunity for me personally to learn
more about the mechanisms and the way everything is set up.
The gripper unit was taken off the X-Y slide to be able to work on it on
the bench.
Gripper was separated to get acces to all mechanisms.
On disassembly I found out the following:
*The O-rings on the piston were dried out
*Some hard grease was found, this could explain the phenomena we once
saw that the gripper jaws don’t open properly (the vent could be blocked by
the hard grease).
*A pully on the theta-axis was slightly bent.
*The limitswitch to prevent the Z axis to run into the mech hardstop was
broken.
Actions:
*All O-rings were changed and lubricated with vacuum grease
*Surfaces on the piston and cilinder were polished to make sure that the
O-rings move smoothly.
*Z limitswitch was changed
*Pully was changed
*The whole unit was cleaned and reassembled.
After this stripdown a series of calibration was needed to make sure
that all the mechanisms were zerosetting correctly.
Theta-axis.
First off all was cheched that it moved freely now. Therefor the
backlash on the worm gear was adjusted. Belt tension was set up to optimum.
Mechanism was checked using engineering terminal.
We resetted the zero position of theta. You do a zeroset and then use
the ruler to measure and see if the gripper jaws are orthogonal to the X and Y
axis. We had to do a small adjustment of 49 encoder counts, this was corrected
in the software.
Tv viewing system.
This system is fitted to the gripper and looks a the fibre end, for
picking and placing is is very important that the offset between centre of theta
rotation and the tv centre is set up correcly.
A fibre was clamped in the gripper, slightly offset. Then theta was
moved through 360 degrees. On the tv screen
the imaged fibre should
describe a circle with its
centrepoint that has to coincide
with the true tv centre. Camera mount was adjusted to achieve this.
Next thing was to check that the camera is orthogonal to the X and Y
axis. This is done viewing a fibre and then
move the robot in one axis X or Y and look on the tv screen if the fibre
moves either horizontal or vertical.
Z-axis
The limitswitch on Z was replaced. The zerosetswitch had to be removed
aswell. Therefore we had to recalibrate this mechanism. This was done with the
fibre module on.
The zerosetswitch was adjusted to make sure that there is enough travel
on the mechanism to reach the outher limit. We measured the distance from the
gripper frame to the gripper jaws with a micrometer. The limitswitch was setup
600 microns further back than the zeroswitch.
Then there’s still another 600 microns before it runs into the
hardlimit.
Sleep |
6.75 |
Wake (zeroset) |
7.26 |
Limitswitch |
7.90 |
Mech hardstop in |
8.50 |
Mech hardstop out |
0.32 |
Next thing was to determine the pickup position. We moved the gripper
towards the fieldplate above fibre 121. Then we made sure that 75 % of the
sharksfin was covered by the jaws. The value was read and changed in software.
We made sure that the jaws were aligned aswell. Some more test picking and
placing fibres were done.
We checked aswell that the fibre is lifted high enough to clear any
other fibres. We’ve seen in the past that fibres catch on each other if
they’re not lifted enough.
Z-pickup value |
5500 encodercounts |
During testing we’ve found out that the belt tension was set to tight
and occasionally the mechanism stalled.
Other things discovered while
testing:
*Bad connector on power amplifier crate: refitted with different
powercable.
*Bad connector on power supply in instrument: reseated connector
*Voltage on powersupply to linedriver dropped below 5V: recalibrated
potentiometer.
After this we did several setups, small_circle/small_park at different
instrument orientations.
Work done during run september 2002
First off all to mencion that the two !! S/D nights were wiped out by
bad wheather. Saturday was the first science night and the wheather was better.
I’ll give a summary about the faults that occurred and the actions
taken.
Saturday 1-9
Fault 15294; drama task failed. This is a known thing apart from this I
discovered from the log that it failed twice earlier on a setup due to a failure
on a centroid while putting a fibre down. The next morning I tried to do the
setup that failed and was fine.
Sunday 2-9
Fault 15297; robot failure.
According to the log while doing setup a2589_bright.cfg;1 it failed to
put down fibre 125. So there;s some suspicion on the Z movement. When it failed
to put a fibre down it will retract in Z and sit there with the fibre in the
jaws. It won't allow you to do anything until you release the fibre. Later on
the VME was rebooted what takes the airpressure off, so the fibre will be
released. The gripper status was open now so should be ok to do a wake, still
would not work. This morning we tried to wake from engineering terminal wouldn't
work. Took off module checked all the belts and that motors were servoing. Put
module back on an woke from ICL, no problems. We then investigated the cabling
if that was causing the problem, ok. Spent some time on the Z mechanism,
readjusted the belt tension although it seemed allright. Did a soak test on
waking and sleeping and moving the Z mechanism in and out, no failures. Then did
the same field setup that failed previous night, no problem. Did an allcircle
and setup of the same field again now with the telescope at zenith, no problem
The only thing that we can think off that could have caused this is the Z limit switch. Altough it seems hard to believe that it was operated if you bear in mind that from 0 (encodercounts) it has a 600 micron clearance and the pickup position is 400. That means there 400 * 1,36 = 544 microns plus 600 microns clearance??
Monday 3-9
No faults
Tuesday 4-9
Fault 15299: failure to ftp files: due to software bug O/S9.
Fault 15300:
We tried the same setup as last night when it failed (a376_bright.cfg;1). We managed to reproduce the fault this time on fibre 14. It failed to centroid. We believe this is because it didn't pick up the fibre, the robot then moves to the putdownposition and the tv camera will look at it but won't see it and come up with teh error "centroid failure". We took off the module, investigated if the airpressure could be causing the problem, but with the current pressure gripper operation was ok. Current pressure is 6 bar. From 4 bar and lower the gripper starts to fail. Checked the cabling on the solenoid that operates the gripper, ok. RAB reduced the Z speed and acceleration and this seems to make the mechanism more smooth. After this run we might have to think abouty replacing the leadscrew on this mechanism, it sounded a bit rough with the old (faster) speed. We did another setup (a2169_faint_recent.cfg;1), this time it failed on fibre 148, pickup error, that indicates that it failed to move down in Z. This could be caused by the fact that it moves down and hits the shoulder on the button, that will cause the Z drive to retract to 400 and it will stay there. We then changed the Z value for pickup. It was at 5500 and we changed it to 5375 which means that there's 170 microns more clearance between the gripper jaws and the button shoulder. Later on we saw that it failed to view a fibre before pickup when it was actually visible on the tv monitor. We checked all the cabling on the VME rack for any loose connections. We changed one of the DMC cards that had an error light on it. Cooling on the VME was improoved. Finally we did a last setup (a2169_faint_recent) and that completed ok. Maybe a matter of overheating of the VME?
Wednesday 5-9
No more faults. Occasionally D task lost. The intervals of this failing got larger. This probably due to the fact that KMD properly terminated the loose end on the network cable at UES.
Thursday 6-9
Ok
Friday 7-9
Ok
Saturday 8-9
Ok
Sunday 9-9
Ok
Conclusion.
It still seems hard to understand what happened on Sunday night when we lost 6 hours. The only thing we can think of is the Z limit switch. Again as I explained above it seems weird, but maybe with the higher speed that we had before it over travelled? Maybe we have to talk about getting rid of this switch?
I think the other problems that occurred were very random and therefore is difficult to try and find a relation. It was interesting to see that as soon as we changed the cooling on the VME that we didn’t have any more problems. The room where the electronics sit, is v ery full now with all the racks of Wyffos, Naomi, Integral and the Ingrid compressor. The latter is causing a big heatload on all of the electronics. This situation with the coming of Grace will be of the past.
Further we still had the D-task dropping of the network from time to time, what doesn’t make things easier to work. In the end it all contributes to a general feeling of frustration. The same thing for all these error messages you get on the ICL. We, staff know not to pay to much attention to them, but an observer might get worried and it certainly doesn’t help to make them feel confident about the instrument.
Finally it’s a real shame that we lost the two S/D nights to bad wheather, because I think we probably would have caught some problems in an earlier stage. Especially since Sunday was their first night on their own and was when they really were doing a lot of setups.
Michiel van der Hoeven AF2 instrument manager